This model is a clone of the powerful instruction-following model, Gemma 3 270M IT, originally developed by Google. This repository serves as a personal mirror for testing and development under the username Maazwaheed.

Model Details

Model Description

The original model is a lightweight, decoder-only Large Language Model (LLM) designed for instruction-following and dialogue. It is a smaller variant of the Gemma family, fine-tuned specifically for chat and interactive use cases.

Developed by: Google (Original)
Shared by: Maaz Waheed (Maazwaheed)
Model type: Causal Language Model (LLM), Decoder-Only Transformer
Language(s) (NLP): English
License: The usage is governed by the original Gemma license provided by Google.
Finetuned from model: google/gemma-3-270m-it

Model Sources

Original Model Source: Google's Gemma Repository (Hypothetical Link)
GitHub: https://github.com/42Wor

Uses

Direct Use

This model is intended for:

Instruction Following: Responding accurately to user prompts.
Simple Chat: Engaging in multi-turn dialogue.
Experimental Development: Being integrated into small applications where low latency and small size are critical.

Out-of-Scope Use

The model is not intended for generating hate speech, malicious code, or engaging in high-stakes decisions without human oversight. Its small size means it may struggle with complex reasoning or long-context tasks compared to larger models.

How to Get Started with the Model

You can load and use this model directly using the Hugging Face transformers library.

python

import torch
from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "Maazwaheed/mallow-gemma-3-270m-it"

# Load the tokenizer
tokenizer = AutoTokenizer.from_pretrained(model_id)

# Load the model
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    torch_dtype=torch.bfloat16, # Use bfloat16 for better performance/memory
    device_map="auto"
)

# Example Chat Interaction
chat = [
    {"role": "user", "content": "Explain the concept of quantum entanglement in simple terms."},
]

prompt = tokenizer.apply_chat_template(chat, tokenize=False, add_generation_prompt=True)

inputs = tokenizer.encode(prompt, add_special_tokens=True, return_tensors="pt").to(model.device)
outputs = model.generate(inputs, max_new_tokens=100)

print(tokenizer.decode(outputs[0]))